Can't use pre-trained data frozen_inference_graph.pbsegmentaa, ab, ac and ad
Closed this issue · 3 comments
Hi,
Thank you for your help on the previous issue. I successfully go through all the steps, but the output images don't contain any identified objects. To figure out the problem I'm trying to use your pre-trained data but analyzeimages.py can't read them because it's separated as frozen_inference_graph.pbsegmentaa, frozen_inference_graph.pbsegmentab, frozen_inference_graph.pbsegmentac and frozen_inference_graph.pbsegmentad. Would you please let me know how to use them with analyzeimages.py?
Edit: I just found the detection scores are pretty low with my training data which has more than 20,000 steps. So it looks like my training has some problem. I used the sample images in this git. Any comments about the problem?
Thank you.
Hello,
I split apart the model because Github has a file size limit. You can concatentate the files back together, for instance something like cat frozen_inference_graph.pbsegment?? > frozen_inference_graph.pb
What was your Total Loss value after 20,000 steps? Were you running tensorboard during the training?
Thank you for your reply, attached is my loss graph and evaluation images. I ran the tensorboard during all my training. The loss value is pretty low but the diagram looks strange. The evaluation images detect everything as the bicycle. I just tried your pre-trained model and the result is pretty good.
Those images are from step 0. Tensorboard should rerun them for further steps. I dont know why you arent seeing eval images for later steps. You could try just exporting a frozen version of your model and running the analyzeimages to see the results.
I am glad you are seeing good results with the frozen model I posted.